Welcome!
After the wonderful introduction to the leaflet package,
it is time to put all this knowledge into practice. We will show you how
to:
dplyrWe are going to create an election map using the features learned in the presentation. This is going to be a replication of the presidential election in 2016 in Pennsylvania.
First things first. We need to lead the leaflet package with other few ones that will make our task easier. Also, we need to get the data and the base map.
library(leaflet)
library(tidyverse)
library(htmltools)
library(rgdal)
library(sf)
library(scales)
library(glue)
# Importing the data set
pa_election_2016 <- read_csv("pa_2016_presidential.csv")
head(pa_election_2016, 5)# Importing the base map
pa_map <- sf::st_read("PaCounty2020_08/PaCounty2020_08.shp",
stringsAsFactors = FALSE)FALSE Reading layer `PaCounty2020_08' from data source
FALSE `D:\DANIEL\MPP Hertie\Year 2 - 1st semester\Intro to Data Science\DS Workshop\Exercise\PaCounty2020_08\PaCounty2020_08.shp'
FALSE using driver `ESRI Shapefile'
FALSE Simple feature collection with 67 features and 25 fields
FALSE Geometry type: MULTIPOLYGON
FALSE Dimension: XY
FALSE Bounding box: xmin: -81 ymin: 40 xmax: -75 ymax: 42
FALSE Geodetic CRS: NAD83Nothing out of the ordinary, isn’t it? Well…
With the power of dplyr, we should check if the county
names coincide in both files, and then change it to prepare the data for
the interactive map.
First change the column names to match both data sets:
pa_map <- pa_map %>%
rename(County = COUNTY_NAM)
head(pa_map, 5)Are the county names exactly the same in both datasets? Let’s check it out!
problems <- anti_join(pa_map, pa_election_2016, by = "County")
head(problems[,1:3])This county is written in ful upper cases in the base map, but as “McKEAN” in the elections dataset.
Let’s change the name in the election dataset!
pa_election_2016$County <- recode(pa_election_2016$County,
McKEAN = "MCKEAN")
# Check again for inconsistencies in names
anti_join(pa_map, pa_election_2016, by = "County")## Warning in `[<-.data.frame`(`*tmp*`, is_list, value = list(`26` =
## "<s_GEOMET>")): replacement element 1 has 1 row to replace 0 rowsWith the county names checked, it is time to merge the base map and the electoral data set into one. Also, we must do some final adjustments.
pa_map_2016 <- full_join(pa_map, pa_election_2016, by = "County")
# Change the turnout proportion of each candidate to percentages
pa_map_2016 <- pa_map_2016 %>%
mutate(TrumpPct = TrumpPct*100) %>%
mutate(ClintonPct = ClintonPct*100)
# Change the projection to match the one leaflet tiles uses (WGS84)
pa_map_2016 <- st_transform(pa_map_2016, "+proj=longlat +datum=WGS84")Let’s map the percentage differences between Trump and Clinton in each county. As it is more complicated to include negative values on the map, we will play with the positive margins for each candidate separately and then merge it into the same map.
Step 1: Create a palette for each candidate
trump_palette <- colorNumeric(palette = "Reds",
domain = pa_map_2016$PctMargin)
clinton_palette <- colorNumeric(palette = "Blues",
domain = pa_map_2016$PctMargin)Step 2: create two separated data frames, one for each candidate.
trump_df <- pa_map_2016 %>%
filter(Winner == "Trump")
clinton_df <- pa_map_2016 %>%
filter(Winner == "Clinton")Step 3: create the pop-ups with the corresponding labels.
We will use the htmltools package to create the pop-ups
for each candidate.
trump_popup <- glue("<strong>{trump_df$County} COUNTY</strong><br />
<strong>Winner: Trump</strong><br />
Trump: {scales::comma(trump_df$Trump, accuracy = 1)}<br />
TrumpPct: {scales::comma(trump_df$TrumpPct, accuracy = 1)}<br />
Clinton: {scales::comma(trump_df$Clinton, accuracy = 1)}<br />
ClintonPct: {scales::comma(trump_df$ClintonPct, accuracy = 1)}<br />
Margin: {scales::comma(trump_df$PctMargin, accuracy = 1)}") %>%
lapply(htmltools::HTML)
clinton_popup <- glue("<strong>{clinton_df$County} COUNTY</strong><br />
<strong>Winner: Clinton</strong><br />
Clinton: {scales::comma(clinton_df$Clinton, accuracy = 1)}<br />
ClintonPct: {scales::comma(clinton_df$ClintonPct, accuracy = 1)}<br />
Trump: {scales::comma(clinton_df$Trump, accuracy = 1)}<br />
TrumpPct: {scales::comma(clinton_df$TrumpPct, accuracy = 1)}<br />
Margin: {scales::comma(clinton_df$PctMargin, accuracy = 1)}") %>%
lapply(htmltools::HTML)Step 4: create the map by adding the polygons for each candidate.
leaflet() %>%
addProviderTiles("OpenTopoMap") %>%
addPolygons(
data = trump_df,
fillColor = ~trump_palette(trump_df$PctMargin),
label = trump_popup,
stroke = TRUE,
smoothFactor = 0.2,
fillOpacity = 0.8,
color = "#666",
weight = 1
) %>%
addPolygons(
data = clinton_df,
fillColor = ~clinton_palette(clinton_df$PctMargin),
label = clinton_popup,
stroke = TRUE,
smoothFactor = 0.2,
fillOpacity = 0.8,
color = "#666",
weight = 1
)Disclaimer: Only downside with this method: it is almost impossible to add a legend as the addLegend function cannot identify simultaneously values from two different data frames.
Now that you know how to do cool interactive maps, it is your turn to create your map with the data and base maps we provided you in this workshop. Keep it simple and have fun!
# PUT YOUR CODE HERE#Sources
This tutorial drew heavily and was adapted from this demo by Sharon Machlis